Several interventions have been evaluated in individuals with newly diagnosed type 1 diabetes, in an attempt to interdict the disease process and preserve B-cell function. Interpretation of these has been complicated by a number of factors. Early studies often used 'remission' as an outcome, based on cessation of insulin therapy or very low doses of insulin. In fact, there was even a recommended definition of remission promulgated52. Yet, most recent investigations have focused more on preservation of C-peptide as a biochemical marker of B-cell function2. Moreover, it has come to be appreciated that more intensive insulin therapy and/or better maintenance of glycaemic control results in better preservation of B-cell function53-56. Of these, probably the best data come from the Diabetes Control and Complications Trial (DCCT)56. Individuals who entered the DCCT with high residual B-cell function (stimulated C-peptide levels of 0.2-0.5 pmol/ml) (n=303) and who were randomized to intensive therapy to maintain normoglycemia, showed a significantly slower decline in stimulated C-peptide levels than did those randomized to conventional therapy56. As a consequence of these observations, aggressive glycaemic control with intensive insulin therapy is important for preservation of endogenous B-cell function in patients with type 1 diabetes. Current strategy involves maintaining insulin therapy at the highest dose that does not induce hypoglycaemia. Moreover, this precludes use of remission characterized by cessation of insulin therapy or reduction of insulin dose as an end-point in clinical trials. The current approach, used by most investigators in the field, and consistent with the recommendations of the Immunology of Diabetes Society57, is to use C-peptide response to evaluate B-cell function.
A wide variety of interventions has been considered, and many have been tested in small pilot studies or in clinical trials2-13. In this chapter, discussion will generally be confined to those interventions that have been tested in randomized, controlled clinical trials.
Cyclosporine. Cyclosporine has been studied in new onset diabetes in several randomized trials, as well as a number of open trials58-67. Two of these - the Canadian/European Study60 and the French Cyclosporine Study58 - were large multicentre randomized controlled trials with sufficient subjects (188 and 122 subjects, respectively) to draw meaningful conclusions. These (as well as the other smaller studies) have demonstrated that cyclosporine results in either better preservation of B-cell function and/or greater likelihood of remission than that seen in placebo or historical control patients. Although at least one study suggested that there was a sustained beneficial effect after discontinuation of cyclosporine66, the authors concluded that the magnitude and duration of that benefit do not appear sufficient to justify cyclosporine treatment in clinical practice. This is particularly the case given the potential of nephrotoxicity with cyclosporine usage68,69. The toxicity of cyclosporine is such that it cannot safely be considered in new-onset type 1 diabetes. Cyclosporine studies included the first two large-scale randomized controlled trials of immune intervention in type 1 diabetes. Moreover, they were important in that they convincingly demonstrated that the type 1 diabetes disease process could be impacted by immune intervention, thus fulfilling the criteria that human type 1 diabetes is an immune-mediated disease.
Azathioprine. Azathioprine has been studied in several small randomized trials in new onset diabetes, either alone or in combination with glucocorti-coids70-73. As with cyclosporine, in some - but not all - of the studies, subjects treated with azathioprine had more insulin-free 'remissions', lower insulin requirement, and higher glucagon stimulated C-peptide levels. One exception was a study in children aged 2-20. Azathioprine may result in severe leukopenia (but this can be avoided with careful monitoring and dose titration) and there is an unquantified concern about oncogenic potential. As a consequence of fears about potential toxicity, there were no further studies of azathioprine in type 1 diabetes.
Nicotinamide. Nicotinamide has been used in many studies in new onset diabetes74-90. Results of individual studies have been mixed, with some showing marginal beneficial effects of nicotinamide and others being without effect. To collectively examine the potential effects of nicotinamide, a meta-analysis of the first 10 of these studies has been reported91. The 10 randomized controlled trials (five of which involved placebo) conducted in recent-onset diabetes and included in the meta-analysis involved a total of 211 nicotinamide-treated patients. One year after diagnosis, baseline C-peptide was significantly higher in nicotinamide-treated patients compared with control patients (0.73 ± 0.65 versus 0.32 ± 0.56 ng/ml, p < 0.005). Moreover, the statistical difference remained when the analysis was confined to the five placebo-controlled trials. This combined analysis suggests a therapeutic effect of nicotinamide in preserving residual B-cell function when given at diabetes diagnosis in addition to insulin. Adverse effects of nicotinamide were minimal. Thus, it is surprising that there has not been much apparent use of nicotinamide in new onset diabetes, although it is being studied in a large prevention trial, as discussed below.
17 subjects appeared promising, in that it led to transient 'clinical remission'92. However, three randomized controlled trials (involving a total of 192 subjects) of BCG in new-onset diabetes failed to demonstrate a beneficial effect after one to two years93-95. Although this would apparently put to rest the notion of using the BCG vaccine, unfortunately these studies may suffer from each being too small and thus too under-powered to firmly conclude that there was no effect.
Linomide. Linomide (quinoline-3-carboxamide) is a synthetic immunomodulator which results in complete protection from insulitis and diabetes in NOD mice96. The effects of linomide on insulin needs and B-cell function were evaluated in a one-year study in 63 subjects (randomized 2:1 linomide and placebo) with recent onset diabetes97. In the linomide group, both insulin dose and HbA1c were lower, and there was a trend for higher C-peptide values. In a post-hoc sub-group analysis performed in 40 patients (25 from the linomide group and 15 from the placebo group) who still had detectable residual B-cell function at entry, linomide was associated with higher C-peptide values. The authors felt that these results support further studies to define the effects of linomide in type 1 diabetes. Unfortunately, the manufacturer withdrew linomide from further study due to side-effects in other (non-diabetes) trials. A similar drug, leflunomide, has been marketed for the treatment of rheumatoid arthritis, but has not been evaluated for potential effects in type 1 diabetes.
Insulin. As noted earlier, several studies have suggested that early and more aggressive insulin treatment may result in preservation of B-cell function, better metabolic control, and/or a prolonged honeymoon period53-56. As a consequence, at least two studies were specifically designed to evaluate the effects of vigorous insulin therapy on preservation of B-cell function in recent-onset diabetes. In a study from Tampa, 26 adolescent subjects were randomized either to conventional insulin therapy or to a two-week course of intravenous insulin delivered via an artificial pancreas to maintain blood glucose levels in the low normal range98. During the two-week intervention, the experimental-therapy group received four times more insulin than the conventionally treated group. Subsequently, both groups were treated similarly. The experimental group had better preservation of B-cell function (meal stimulated C-peptide levels) and lower HbA1c levels for at least one year after randomization. A study from Munich sought to clarify whether the beneficial effect seen in the Tampa study was a consequence of the intensification of therapy or the route of insulin administration (intravenous)99. To that end, 10 subjects with newly diagnosed diabetes were randomized to either a two-week high-dose intravenous insulin infusion or intensive insulin therapy with four injections per day. By the third week, both groups were treated similarly with intensive insulin therapy and were followed for one year. Changes in stimu lated C-peptide concentrations between months 0 and 12 were not significant in either group, suggesting that both therapies were effective in preserving B-cell function.
Oral insulin. Oral administration of insulin to young NOD mice decreases insulitis and delays the onset of diabetes100-102. This is consistent with the immunological concept of 'oral tolerance'103, which suggests that antigens administered by the oral route favour the generation of the T-helper-2 (Th2) and T-helper-3 (Th3) subsets of CD4+ T-cells and the 'type 2' and 'type 3' cytokines they respectively produce, cytokines which inhibit the T-helper-1 (Th1) and 'type 1' cytokine mediated B-cell destruction which leads to type 1 diabetes. That these regulatory cells have been generated was shown by the fact that co-transfer of spleen cells from animals treated with oral insulin prevents adoptive transfer of diabetes100,103. The idea of using oral insulin to slow the destructive processes in human type 1 diabetes is appealing, since oral insulin has no metabolic effects, and the safety of the insulin molecule in human beings is well established. Three randomized controlled trials (involving a total of 418 subjects) have evaluated oral insulin in new-onset diabetes104-106. Unfortunately, oral insulin did not modify the rate of decline of B-cell function, although one study claimed potential beneficial effects using post-hoc analyses.
Heat-shock protein peptide. The 60 kDa heat-shock protein (hsp60) is thought to be one of the target self antigens involved in the B-cell destruction which leads to type 1 diabetes107. An immunomodulatory peptide from hsp60, p277, arrests B-cell destruction and maintains insulin production in newly diabetic NOD mice108. In a small randomized controlled trial in 35 subjects with diabetes diagnosed within six months, the treated group showed better preservation of C-peptide at 10 months, suggesting a potential beneficial effect109. These results support further studies to define the effects of hsp60, p277 in type 1 diabetes.
Anti-CD3 monoclonal antibody. Studies in mouse models of type 1 diabetes have shown that non-FcR binding anti-CD3 monoclonal antibody can prevent development of insulitis and hyperglycaemia and can even reverse diabetes, and appear to induce a state of tolerance to diabetes110-112. A non-FcR binding anti-CD3 monoclonal antibody [hOKT3g1(Ala-Ala)] was studied in a randomized controlled phase I/II trial in 24 subjects with new-onset type 1 diabetes113,114. Treatment involved a 14-day course of the antibody in escalating doses for four days followed by full dose (45 ug/kg/d) for 10 days. The treated group showed improved glycaemic control and better preservation of C-peptide (in response to a mixed meal) at 12 months, with the suggestion that the treatment effect was maintained at 18 months as well. Thus, this very small study suggested a beneficial effect, leading to the conclusion that this therapy warrants further evaluation in larger randomized controlled trials.
Studies in Individuals at High Risk of Type 1 Diabetes
In addition to attempts to interdict the disease process and preserve B-cell function in newly diagnosed type 1 diabetes, studies have been initiated to try to delay or prevent the clinical onset of type 1 diabetes. Although a number of immunosuppressive drugs (e.g. azathioprine or cyclosporine) have been considered and some even given to a few individuals115118, these were not really evaluated in a disciplined way. More recently, large-scale multicentre randomized controlled clinical trials have been initiated to evaluate nicotin-amide, parenteral insulin, oral insulin and the elimination of cows' milk from infant feeding.
Nicotinamide. Nicotinamide has been used in prediabetes. No effect was seen in one tiny pilot study119. In two others, subjects given nicotinamide appeared to fare better than untreated historical control subjects120. In another study, insulin secretion seemed better preserved in six treated subjects than in seven control subjects121.
The largest nicotinamide study to date emanates from Auckland, New Zealand122. In this study, during the period 1988-91, school children aged 5-8 (with no immediate family history of diabetes) were randomized by school to receive ICA testing. A total of 33658 children were offered testing: 20195 accepted and 13 463 declined. Of those tested, 185 were ICA positive. Of these, 173 were treated with nicotinamide (maximum dose 1.5 g/day) on the basis of either ICA levels of >10 JDF units and first phase insulin release < 25th percentile of normal, or those with ICA >20 JDF units. Another 48 335 children were neither screened nor treated, and served as controls. They were followed for 7.1 years. The rate of development of diabetes was 7.14/105 per year in the nicotinamide treated group versus 16.07/105 per year in the comparison group. The rate in those who refused testing was 18.48/105 per year. After age adjustment, the tested group had a rate of diabetes of 41% (20-85 95% confidence interval) of the other groups combined, which is significant (p = 0.008). When an intention to treat analysis was performed by combining the treated group with those who refused testing, the rate of diabetes was less than in the comparison group, but the difference did not reach statistical significance. No adverse effects were seen in treated subjects.
Two large multicentre randomized, double-masked, controlled clinical trials have been initiated to evaluate the effects of nicotinamide in high-risk relatives of individuals with type 1 diabetes. These are the German (Deutsch) Nicotinamide Diabetes Intervention Study (DENIS) and the European Nicotinamide Diabetes Intervention Trial (ENDIT).
In DENIS, individuals at high risk for developing type 1 diabetes within three years were identified by screening siblings (aged 3-12) of patients with diabetes for the presence of high titre ICA123. Subjects (n = 55 were randomized into placebo and high-dose nicotinamide-slow release (1.2 g/m2/day) groups and followed in a controlled clinical trial using a sequential design. Rates of diabetes onset were similar in both groups throughout the observation period (maximum 3.8 years, median 2.1 years). The authors assert that the sequential design provides a 10% probability of a type II error against a reduction of the cumulative diabetes incidence at three years from 30 to 6% by nicotinamide. The trial was terminated when the second sequential interim analysis after the eleventh case of diabetes showed that the trial had failed to detect a reduction of the cumulative diabetes incidence at three years from 30 to 6% (p = 0.97). The data do not exclude the possibility of a weaker, but potentially meaningful, risk reduction in this cohort, or a major clinical effect of nicotinamide in individuals with less risk of progression to type 1 diabetes than those studied.
ENDIT is a prospective, placebo-controlled double-blind trial, in ICA-positive, first-degree relatives, aged 5-40 years, of an individual who developed type 1 diabetes under the age of 20 years, and who are themselves positive for ICA47,48. Untreated subjects with these enrolment criteria have a projected risk of type 1 diabetes of 40% over a five-year period. Subjects (n = 528) were randomized into placebo and high-dose nicotinamide (1.3 g/m2/day) groups to determine if a 35% reduction in incidence of type 1 diabetes can be achieved during five years of treatment. ENDIT is an international study conducted in 20 countries. More than 35000 first-degree relatives were screened. Masked comparisons are performed at six-month intervals by a review committee, with study completion projected for 2002.
Insulin. Insulin has been shown to delay the development of diabetes and insulitis in animal models of spontaneous diabetes (BB rats and NOD mice)126-132. In human beings, several pilot studies of insulin use in high-risk relatives of individuals with type 1 diabetes were conducted, two of which have been reported in detail. In one pilot study, from Boston, insulin was offered to 12 subjects, five of whom accepted and seven declined, and served as a comparison group133. In this study, insulin therapy consisted of five days of continuous intravenous insulin every nine months, coupled with twice daily subcutaneous insulin. Life table analysis suggested that this treatment may delay the appearance of diabetes.
In another pilot study, from Munich (the Schwabing Insulin Prophylaxis Trial), there were 14 high-risk first-degree relatives of people with type 1 diabetes randomized to either experimental treatment or a control group134,135. In the experimental treatment group, intravenous insulin was given by continuous infusion for seven days, followed by daily injections for six months.
Intravenous insulin infusions were repeated every 12 months. In the treatment group three of the seven individuals (follow-up from time of eligibility: 2.3 to 7.1 years) and in the control group six of the seven untreated individuals (1.7 to 7.1 years) developed clinical diabetes. Life table analysis showed that clinical onset of type 1 diabetes was delayed in experimental subjects compared with control subjects (diabetes-free survival: 5.0 ± 0.9 years versus 2.3 ± 0.7 years; p < 0.03).
The preliminary results from these pilot studies suggest that in high-risk relatives insulin has the potential to delay or prevent the development of overt diabetes. It was also appreciated that insulin is B-cell specific, does not have generalized effects on the immune system, has well understood effects on people, and has known side-effects that are controllable. This led to the Diabetes Prevention Trial of Type 1 Diabetes (DPT-1), a randomized, controlled, multicentre clinical trial, conducted throughout the USA and Canada to test whether intervention with insulin can delay the appearance of overt clinical diabetes136. DPT-1 screened and analyzed 84 228 samples from relatives of patients with type 1 diabetes for islet cell antibodies (ICA), and found 3152 (3.73%) relatives who were ICA positive on initial testing. Of these, 2103 (67%) underwent staging to quantify risk of type 1 diabetes, and 372 relatives progressed in their staging evaluations to be classified as having a risk projection for type 1 diabetes of greater than 50%. A total of 339 participants were randomized either to the intervention group or to the close observation group. The experimental intervention group received parenteral insulin - both annual intravenous insulin infusions and twice-daily low-dose subcutaneous insulin injections (0.25 units per kg per day). An oral glucose tolerancetest was performed every six months; the diagnosis of diabetes was confirmed by a second test. The median duration of follow-up was 3.7 years. Diabetes was diagnosed in 139 participants - 69 in the intervention group and 70 in the close observation group. The average proportion of participants who progressed to diabetes was 15.1% per year in the intervention group and 14.6% per year in the close observation group. The cumulative incidence of diabetes in the intervention group was virtually the same as in the observation group. Thus, insulin in the dose and regimen used did not delay or prevent the development of type 1 diabetes. However, most participants diagnosed with diabetes were asymptomatic at the time of diagnosis.
Oral insulin. As noted earlier, oral administration of insulin100-101 or insulin B-chain102 to young NOD mice decreases insulitis and delays the onset of diabetes. Moreover, spleen cells from animals treated with oral insulin prevent adoptive transfer of diabetes100,103. Therefore, DPT-1 also included a protocol testing whether oral insulin (7.5 mg/day) can delay the appearance of overt clinical diabetes. In this randomized, placebo-controlled, double-masked, multi-centre clinical trial, nearly 100 000 relatives have been screened to identify and randomize 350 to 360 relatives classified as having a risk projection for type 1 diabetes of 26-50% over five years. The trial will complete enrolment in 2002 and will follow subjects for two years thereafter.
In addition to oral insulin, nasal insulin may also lead to tolerance. Two studies are currently examining its effects. One is a double-blind, placebo-controlled, pilot crossover study in Melbourne, Australia - the Intranasal insulin trial (INIT)137. That study, amongst first-degree relatives, is examining the effects of intranasal insulin on surrogate markers - autoantibodies and T-cell proliferative and cytokine responses to relevant antigens. The other - the Diabetes Prediction and Prevention (DIPP) Project - is a study being conducted in Turku, Finland among newborns from the general population (i.e. without relatives with diabetes) with high-risk genotypes for type 1 diabetes138.
Milk proteins. In some epidemiological and case-control studies, it has been suggested that there is a reciprocal relationship between infant breast-feeding and subsequent development of type 1 diabetes139,140. It has been proposed that breast-feeding may be a surrogate for the absence of consumption of cow milk proteins (CMP)140. These highly controversial hypotheses are supported by one meta-analysis140, but challenged in another as being confounded by bias141. In Finland, a small prospective study suggests that exclusive breastfeeding reduces the risk of diabetes development142. On the other hand, a prospective study in the USA, using appearance of antibodies as the endpoint, failed to find a relationship143. Another Finnish cohort of high-risk newborns found that short duration of breast-feeding, together with early introduction of cow milk proteins, led to increased appearance of islet auto-antibodies144.
In spite of the controversy, the notion has been developed that consumption of CMP, particularly during a 'critical window of vulnerability' early in life, may lead to the initiation of the immunologic attack against pancreatic islet (3-cells and increase susceptibility to type 1 diabetes145. Others argue that the issue really relates to the immune function of the mucosal barrier146. The champions of the cows' milk hypothesis cite an array of evidence in support of the CMP hypothesis - epidemiologic data, disease rates in animal models, humoral and cellular immune markers directed against CMP in patients with new-onset type 1 diabetes, and identification of a peptide sequence on bovine serum albumin (BSA) with homology to sequence on the islet cell protein ICA-69 (or p69) ('molecular mimicry')145. To test the hypothesis, a multinational randomized prospective trial, TRIGR (Trial to Reduce Incidence of Diabetes in Genetically at Risk), has been initiated to determine whether the frequency of type 1 diabetes can be reduced by preventing exposure to CMP during early life147. This randomized prospective trial may screen 6000 infants who have a parent or sibling with type 1 diabetes to identify 2000 newborns who will be followed for 10 years for the development of type 1 diabetes. For eight months they will receive either a conventional CMP formula or a formula in which there has been replacement of CMP with casein hydrolysate. This would be a 'true' primary prevention strategy.
Was this article helpful?